It is known for images which are created by one or more cameras to be used to automatically determine the position of an object in a working area of a robot. In this case, images of the working area from different perspectives are also used. This can make it possible to view the working area three-dimensionally, without restrictions and for example, to also view a part of the working area of the robot which can be seen only from one specific perspective, using partially redundant image information. An effort for determining the position of an object which is located in the working area of the robot using the image data from one or more images can be significantly reduced with the simultaneous presence of the associated coordinates and orientation of the respective camera at the time at which the respective image was created.
The position of an object can also be identified in three dimensions on the basis of a single image, if various constraints are included. For example, the knowledge of a plane on which a flat object, which is not concealed by further objects, is located. However, this represents a practical situation which may occur rarely.
Not only is there a known arrangement of one or more cameras installed fixed in the area surrounding the robot, which cameras create images of a working area of the robot from different, unchanging perspectives but variants are also known which have one or more cameras mounted on the robot.
In the case of an automatic robot-based system, which automatically identifies the position of a loose object located in the working area of the robot, then grips the object with a gripping tool mounted on the robot and transports it to a different point in the working area of the robot, it is known to arrange a single camera on the robot, for example on its wrist joint. One such system is disclosed, for example, in DE 102006059781.8.
Because of the initially unknown position of the loose object in three dimensions, images of this object from specific perspectives can be unsuitable for determining its position, particularly when a plurality of such objects are poured one on top of the other and/or parts of the robot conceal the object for one camera's perspective. A system as mentioned above with a camera arranged on the robot, makes it possible to create images from a multiplicity of individually selectable perspectives, which is not possible when cameras are arranged in fixed positions in three dimensions. Images which are not suitable for determining position can thus be replaced by further images, which are suitable for this purpose.
When creating a plurality of images from different perspectives with a system such as this, the robot can in each case move the camera to the desired positions. From there, the respective images can be created with the robot stationary, as a result of which only one camera mounted on the robot is used, in contrast to permanently installed cameras.
On the other hand, the time for visually identifying the position of an object in three dimensions using image data from one camera attached to a robot can be correspondingly increased because the robot is stationary in different positions.